The recent evolution of magnetic storage technology has included dramatic improvements in magnetic storage medium density and data access times. In order to service the increased density and shorter access times, magnetic recording heads are being called upon to perform their writing and reading functions, as well as their servo signal reading functions, more rapidly. This in turn requires closer timing between successive data transfer operations, or WRITE cycles and READ cycles (both data and servo position signal), which puts stringent requirements on the rapid magnetic relaxation of the magnetic recording head.
However, in prior art magnetic recording systems such as that shown in FIGS. 1 and 2, the relaxation of the magnetic recording head after a WRITE cycle is a highly hysteretic, magnetically irreversible process. This is because when the WRITE ENABLE input deactivates the WRITE DRIVER and thereby abruptly removes the WRITE current from the magnetic recording head, metastable magnetic domain patterns (i.e., marginally stable magnetic patterns) are created in the head which can transition to lower energy states (i.e. relax further) at random points in time after termination of the WRITE process. The associated abrupt transitions in the magnetization of the magnetic recording head results in flux changes threading the magnetic recording head and thereby creating voltage glitches in the, head output. These relaxation glitches, known as Barkhausen noise, can occur for several hundred microseconds following a WRITE cycle, and cause errors in any data or servo signal read during that time.
Additionally, the domain walls, which define the boundaries between the domain patterns, form at different locations and lead to a different, nonzero, final remanence of the head every time the WRITE current is removed from the magnetic recording head. Thus, not only is a given final micro-magnetic domain pattern not reproducible from one WRITE cycle to the next, but the final remanence state of the magnetic recording head is nonzero. As a result, prior art magnetic recording systems exhibit non-reproducible data and servo read performance (due to randomness of the final micromagnetic domain pattern), impaired READ sensitivity following a WRITE cycle (due to nonzero final remanence), and noise-after-WRITE which can produce errors in data and servo READ cycles immediately following a WRITE cycle.
All of the adverse magnetic relaxation effects described above occur in ferrite inductive heads, thin film inductive heads, and thin-film inductive-magnetoresistive heads. Moreover, while observable in standard dual element READ/WRITE magnetic heads, the problems become particularly acute in the single element READ/WRITE magnetic heads that must be used to achieve the alignment accuracy concomitant to increased storage density. Thus, the described magnetic relaxation effects have heretofore detracted from the performance of READ/WRITE magnetic heads and have presented a barrier to reduction of the transition time between WRITE and READ cycles.